The Reliable Analysis Of Distributed Clustered Data

The Obama government in March 29,2012 announced the launch of a big data research and development program, from the information society into the era of big data. In the rapid development of society, science and technology is more and more developed, the flow of information is more and more high, the communication between people becomes more and more close, to give life to bring the unlimited convenience, high data is the product of the high-tech era.In this era of big data, the safety and reliability of the data storage system has become particularly important. The big data amount of data is very huge, and the huge database is composed of structured data and unstructured data, which makes the traditional relational database difficult of big data amount of data for storage, management and processing. It can improve the security of data when the data is distributed in a number of databases.This paper mainly discusses the reliability of distributed cluster data system. A novel semi-Markov process model based on three-model redundant (TMR) architecture is proposed.The semi-Markov process mode is established under the condition that the natural life of each unit of the system is assumed to confirm with a negative exponential distribution and repair time of each unit with generalized arbitrary distribution where the repair rate is considered as a function of time. Simultaneously,the nove lmodel is converted to Markov processmodel by supplementary variable, and then some universal reliability and availability indices of the system model are obtained.The semi-Markov process mode plays an important role in the redundancy structure of multi storage fusion system.Firstly, put forward the important role and the significance of the multi storage memory in life.Secondly, discussing semi-Markov process model is feasibleto the three fusion syste.In the distributed cluster data system model, the system can work normally if there are more than half of the normal working of the storage path. As a result, the model for the reliability of the data system is (N,N/2+1) parallel structure. In a redundant system consisting of this storage path, the system can work normally when there are N/2+1 or more than N/2+l of the stored paths. The failure rate of a single storage path can be regarded as a constant for the multiple storage memory fusion system of the TMR structure which is composed of 3 identical and mutually independent storage path units.When three path also work, a single storage failure rate of λ= 1.0x10-4h, repair time follow arbitrary distribution, the probability density function is g(t)= a2texp(-at),α= 0.5, decomposition in detail and finally the use of mathematical analysis method of distributed cluster data system effectiveness and reliability of that system mean time to failure to be:Then,using mathematical to analysis the effectiveness of system.Finally,using the parameters of reference establishing and soving the state equation about system reliability.The average time of failure of a single storage memor is tMTBF=1/λ1=l/10-4=10000h.and the average time of failure of the distributed cluster data system is tMTBF=5-3g(2λ)/6λ[1-g(2λ)]=1.67×106h,Furthermore mean time between fail(MTBF)of the TMR system employing proposed model for the multi-sensor fusion two order of magnitude increases over the single sensor system.The semi-Markov process model illustrates more accurately the operating characteristics of the system and provides theoretical basis to evaluate effectivly the reliability of the multi-sensor fusion system.